The mechanisms underlying cardiac resynchronization therapy have consistently been studied at rest and remain ill defined. Peak stress total isovolumic time (t-IVT) is a major determinant of cardiac output (CO) in chronic heart failure. In this study, pharmacologic stress was used to assess the effects of atrioventricular (AV) delay shortening and ventricular resynchronization elements of cardiac resynchronization therapy. Thirty patients undergoing cardiac resynchronization therapy were studied <6 months after implantation. t-IVT and CO were measured during native activation (left bundle branch block), AV delay shortening (right ventricular dual-chamber pacing), and full resynchronization (atrio-biventricular pacing). Full resynchronization shortened peak stress t-IVT by 9.4 ± 6.2 s/min (p <0.001) and increased peak stress CO by 0.9 ± 0.4 L/min (p <0.001), with the effects in individual patients showing a large correlation (r = −0.64, p <0.001). In contrast, simple AV delay shortening did not shorten peak stress t-IVT nor increase peak stress CO, nor was CO at rest affected by full resynchronization or AV delay shortening. Of all measurements during native activation, the best predictor of gain in peak stress CO from full resynchronization was peak stress t-IVT (r = 0.75, p <0.001), with every 5 s/min increment in peak stress t-IVT during native activation predicting a 6% gain in peak stress CO. No conventional measures during native activation at rest or during stress (including QRS duration, the Tei index, tissue Doppler intraventricular delay, and t-IVT at rest) added significant additional information. In conclusion, only during stress does resynchronization consistently increase CO. Second, little of this increment in CO is achieved by AV delay shortening alone. Third, under native activation, long t-IVT during peak stress is the single best predictor of resynchronization-mediated increment in peak stress CO.